Numerical Integration Drift for Seismic Acceleration Data

[matthewnelson]

Hello,

I am attempting to perform numerical integration (Newmark-beta method) for acceleration data of a structure from a seismic event to find the velocity and displacement over time. The acceleration data shows a damping sinusoid, oscillating about the x-axis. All initial conditions are set correctly however there is still a linear drift when calculating the velocity leading to an exponential drift in the displacement. I have tried a few other methods as well such as Simpson's rule, a Kalman filter, and basic trapezoidal rule which all resulted in the same issue.

Are there more adaptive procedures to numerically integrate this acceleration dataset in order to obtain accurate velocity and displacement data?


Best,
Matthew